Building startups in the AI era has many unique attributes, but the core challenges rhyme with past disruptions: How do you build a durable company when technology and business models keep being reinvented?
A few companies, like Amazon, have a remarkable track record for identifying major shifts, taking a “first principles” approach, and then executing with clarity and excellence as they build the next wave of innovation.
I recently sat down with Jon Jones, Global Head of Startups at AWS, during one of the nearly 200 events that powered Seattle Tech Week 2025. We gathered to celebrate 30 years of transformation — both for Amazon and for Madrona — and to talk about what lies ahead.
The energy in that room was awesome: founders leveraging AI capabilities in innovative ways, rethinking everything from user experience to infrastructure, and testing new business models. Our conversation was a reminder of something I’ve believed for years: building a lasting company requires genuine curiosity, information triangulation, decisiveness, and a bias toward action.
That’s what makes Seattle special, and why this moment matters. For 30 years, we’ve seen this community anticipate what’s next and do the long-term company-building work to help invent the future. And now a new generation of builders is leading the charge, applying past learnings in a relevant way to the opportunities of what I’m calling the “Reasoning Revolution.”
When we look back, the core lesson is clear: Every revolution redefines advantage and generates value at new abstraction layers. The internet changed distribution and information access. Cloud transformed infrastructure from a fixed cost to a utility and made it elastic. Each shift produced new category leaders and left others behind.
The next wave will be no different. In fact, it will be faster and more pervasive than anything before. AI (and, more specifically, reasoning machines and reasoning flywheels) are already reshaping how products are built, decisions are made, and value is captured.
Technology waves create opportunity. They also raise the bar. So what can founders do today to increase the odds that they’re not just building for this moment, but for the long run? During my conversation at Tech Week, I shared three lessons that have proven durable across waves of innovation, and that matter even more now.
Advantage Compounds
Great companies don’t just ride a transformation wave. They build technologies and solutions that shape the wave and compound advantage over time.
In the early cloud era, AWS didn’t win because it launched a comprehensive platform. It won because they created building blocks that developers needed: more developers led to more services, which attracted more customers, which generated feedback that made the infrastructure better. That flywheel led to customer-inspired innovation year after year and created an enduring moat and over $120 billion in annual revenue.
The same principles apply in the AI era. It will not be enough to enhance a legacy product with AI capabilities – even if you have the incumbent advantages of customers, data, context, and workflows. It’s about building reasoning flywheels where every interaction makes your solution smarter, more valuable, and harder to displace. Companies like Cursor (aka Anysphere) are demonstrating this pattern as they challenge incumbents like GitHub. Think about the data you have access to, the insights you generate, and the actions your product enables, and then close the loop. Data → Insight → Action → New Data. That’s the reasoning flywheel, and it is the foundation of the castle with a modern moat.
As you consider how your company engages this flywheel, I’d encourage you to ask yourself two questions:
-
- Where in your product or platform does value accumulate as more people use it?
- Where does the system get better with each customer engagement cycle?
If you can answer those questions and design for compounding advantage, you’re setting the foundation for something durable.
The Most Important Model Is Still the Business Model
Technology waves change the playing field, but business models determine who captures the value. In the early days of AI, it’s easy to get caught up in technical benchmarks and model performance. But history is clear: great technology alone never wins the market. The companies that endure know who they serve, how they deliver value, and how they capture it sustainably.
The most important AI model is the business model. That means founders today need to spend as much energy on pricing and distribution innovation as they do on model selection. We’ve already seen a wave of usage-based pricing, but in an AI world, what is the unit of value? Tasks completed? Tokens consumed? Hours saved? These aren’t academic questions — they’re existential.
The same goes for go-to-market. In previous waves, companies that embedded distribution into their product (think Slack or Zoom during the SaaS era) accelerated growth dramatically. In this wave, who are your natural distribution partners? How does your product create network effects that pull in more users without linear spend?
Founders who iterate on business models as quickly as they iterate on code will build companies that last. Those who don’t, risk being the “better technology” that never becomes a business.
Learning Beats Certainty
If there’s one constant in every technology wave, it’s uncertainty. That’s why the best founders aren’t the ones with the most confidence; they’re the ones with the deepest curiosity. And, as the rate of change increases, so does the value of learning.
Genuine curiosity is a competitive advantage. It drives you to seek information others overlook, to test assumptions, and to challenge your own biases. But curiosity alone isn’t enough; you need a system for turning insight into action. I call this the learning flywheel: Curiosity → Synthesis → Action → Reflection.
Start by acknowledging what you don’t know. Gather inputs from diverse sources: customers, advisors, competitors, and even adjacent industries. Then synthesize the signals and make a decision, even if it’s imperfect. Because the worst decision is no decision. Every move generates new data, and that data feeds back into your next decision.
The founders who embrace this cycle—who learn faster than the market changes—are the ones who build enduring companies like Amazon. In an era where the pace of change feels relentless, this mindset is not optional. It’s the difference between navigating the wave and being swept away by it.
What Endures
Most companies struggle to build a vision or product that spans more than one wave. Those that do, shape entire industries.
The next era is the Reasoning Revolution. As this era unfolds, advantage won’t come from chasing every trend. It will come from the ability to harness what’s new without losing sight of what endures: clarity of purpose, speed of execution, and systems that compound.
At Madrona, we’ve seen this firsthand: across decades, technical revolutions, and enduring partnerships. Technology waves don’t wait, and we’re here for what’s next.
Looking back at the last era and the current one, the question for every founder is simple and hard: Will you build a company that rides the wave, or one that thrives through the next?